Mistral 3 Family
Mistral·
LLMsflagshipcloud + localFrontier
Overview
Flagship Mistral Large 3 (675B MoE) and efficient Ministral 3 edge models.
Capabilities and innovations
675B MoEMultimodal256k ContextEdge-optimized MinistralNative Multimodal MoEEdge-Cloud Model FamilyApache 2.0 Weights
Benchmarks
No benchmark values are recorded for this model.
- GPQA Diamond: not reported by the vendor
- Humanity’s Last Exam (no tools): not reported by the vendor
- SWE-bench Verified: not reported by the vendor
- SWE-bench Pro: not reported by the vendor
- MMLU: not reported by the vendor
- MMLU-Pro: not reported by the vendor
- Terminal-Bench 2.x: benchmark did not exist at release
- MMMU-Pro: benchmark did not exist at release
Architecture and hardware
- Parameters
- 675B total (MoE)
- Estimated VRAM at Q4
- ~389 GB, Frontier class
Reliability
- Hallucination rate (Vectara HHEM)
- 14.5% (lower is better)independent
- Agentic tool use (τ-bench)
- 70.2% (higher is better)third party
Links
More from Mistral
Data curated by AI Model Timeline. See the methodology for admission criteria, benchmark eras and source priority.